FDA Approves AI Software That Diagnoses Eye Disease

While not in my direct context of focus, it’s an example of technology that will challenge status quo, making its way to all aspects of healthcare.

Diabetic retinopathy is an eye condition that can present in poorly controlled diabetics. If not managed properly, diabetic retinopathy can result in vision loss. Studies have shown that it is the leading cause of vision loss in adults between the ages of 20 and 74, with over one-third of diagnosed diabetics suffering from this condition.

Diabetic retinopathy is typically a diagnosis made by an ophthalmologist, who sees the patient after referral from a primary care provider. This can sometimes result in a delay in diagnosis or treatment as the patient may not be able to see the specialist right away. With diabetic retinopathy being a progressive disease that gets worse over time, it is crucial to minimize any delay in its diagnosis and treatment.

The Food and Drug Administration (FDA) has approved the first ever AI software, called IDx-DR, that can make a diagnosis of diabetic retinopathy in affected individuals without the need to see a specialist. The primary care provider uses a special camera to take pictures of the patient’s retina; the software then analyses the images to establish a diagnosis of diabetic retinopathy. If a positive diagnosis is made, the patient is then referred to a specialist for further treatment.

In a clinical trial with 900 images to establish the efficacy of IDx-DR, a correct positive diagnosis was made 87 percent of the time. It was also able to detect images without diabetic retinopathy 90 percent of the time. The FDA threshold for approval of the software was that it should be able to make a positive diagnosis in at least 85 percent of patients and also detect patients without this condition 83 percent of the time. This threshold was cleared by IDx-DR which was why it was approved by the FDA.

IDx-DR promises to simplify management of diabetic retinopathy as diagnoses can be made much earlier. Furthermore, it may reduce healthcare expenses as it would reduce the need to see a specialist in making a diagnosis of diabetic retinopathy.

Our collective goal is to improve outcomes and reduce cost; AI is going to accelerate our efforts.

Information courtesy of: https://www.linkedin.com/pulse/fda-approves-ai-software-diagnoses-eye-disease-thomas-herzog/